Credit Approval Mechanism with Computer Vision
Input Sources
Images
Businesses & Licenses
Business Impact
70%
improved customer satisfaction
Outcome
83%
improved credit rating model
Problem statement
Customer had a unique scenario where the Ability to get exact information from the ground level executives about loan eligibility in European regions is tough. Most of the Loan defaulters are small and medium segment business owners. Ability and ways to reduce defaulters through Computer Vision & Machine Learning would solve the issue
Challenges
- Leveraging Images of business outlook and the licenses to make sure all the business needs are genuine.
- Processing variety of image data and extracting right information and patterns.
- Integrate the outcome with existing credit scoring model.
- Managing & designing data model for large image corpus.
Technologies used
Solution
- Antz helped customer to design a solution that can integrate with existing model of loan defaulter prediction with Computer Vision based approach.
- Able to create the tags and define the custom vision which can help them to give better prediction for customers approaching on loan.
- Find the users business locations, pictures inside and outside the stores , electronic devices review and many other parameters are considered and taken using pictures and provide an outcomes as JSON which will be passed as an input to existing machine learning model.
Result
- Improved the quality of credit approval mechanism.
- Improved customer acceptance by 15% as most of the customers are not required to meet the banks for loans eligibility.
- Improved customer satisfaction and banks can easily approve the bank loans with confidence.